NEWS & VIEWS CLIMATE SCIENCE
The mysteries of Sahel droughts Past variability in Sahel rainfall is closely linked to global sea surface temperature distributions in observations and models alike. Climate simulations for the 21st century suggest that additional influences may become important in the future.
Kerry H. Cook
2.5
is at the Jackson School of Geosciences, University of Texas at Austin, 1 University Station C1100, Austin, Texas 78712, USA. e-mail:
[email protected]
M
1.5 Precipitation anomalies (mm d–1)
ortality risk from drought is extremely high across subtropical Africa in both hemispheres1. In the Sahel — the transition zone between the tropical climate of the Guinean coast region and the extreme aridity of the Sahara Desert — persistent drought from the early 1960s into the 1980s caused millions of deaths and hampered development. Early concern focused on the influence of land-use practices2, but later observational3 and modelling4–6 studies related decadal-scale Sahel drought with sea surface temperatures. Particularly strong links were found with north–south temperature gradients in the tropical Atlantic and sea surface temperatures in the tropical Pacific and Indian oceans. Higher rainfall rates returned to the Sahel in the 1990s despite a continued warming of the Indian Ocean7, but the population of the region remains vulnerable. Analysing state-of-the-art climate model simulations, Michela Biasutti and co-authors8 found that the simple statistical model linking sea surface temperatures to Sahel rainfall in simulations of the past does not hold for simulated future climate. While preparing the Fourth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC)9, climate modelling groups around the world ran standardized simulations of 20th and 21st century climate. More than 20 groups developed and ran coupled general circulation models (GCMs), that is, models of the global climate that include fully interactive atmospheric and ocean components. These models solve the governing equations of fluid dynamics — the Navier–Stokes equations, appropriately scaled for this application — in three dimensions, covering the globe and integrating forward in time. Horizontal grid spacing ranges from roughly 100 km to 400 km. The governing
2
1 0.5 0 –0.5 –1 –1.5
–2.5
GFDL GISS_E_H MIROC MPI_ECHAM5
MRI NCAR_PCM1 NCAR_CCSM UKMO_HADCM3
1960 1970 1980 1990 2000 2010 2020 2030 2040 2050 2060
2070 2080 2090 2100
Figure 1 Future projections of Sahel rainfall. Time series of Sahel precipitation anomalies for 1960–2100 as simulated by eight coupled general circulation models of the global ocean and atmosphere. Precipitation is averaged over West Africa from 10–20° N and smoothed using a five-year running mean, and anomalies are the differences from the 1950–1999 mean for each model. Biasutti and colleagues7 find a puzzling mismatch in the future evolution of sea surface temperatures and Sahel rainfall in the various models, unlike the simple regression model that explains past variability reasonably well.
equations are accompanied by physical parameterizations to supply processes that are not explicitly treated in each model, such as sub-grid scale diffusion, radiation, cloud formation and precipitation processes. The output from these models has been archived and made available to the scientific community. It is an invaluable resource, allowing extensive model intercomparisons to improve our understanding of present and future climate. A number of researchers have used this archive, or a subset of the full archive, to study Sahel rainfall6,10,11. They have found that many of the 20th century simulations produce strong rainfall in a relatively narrow band near the
equator throughout the summer rather than reproducing the observed rainfall maximum over the African continent that accompanies the development of the West African monsoon11. In simulations of the future, most of the models project only modest increases or decreases in summer precipitation. However, one model — in both a higher and lower resolution version — produces extreme wetting, and another model (again, in two versions) produces extreme drying (Fig. 1). When these outlier simulations are included, the simulated range of future climate outcomes for the Sahel in the model archive is broader than for most other regions. It is not easy to dismiss
nature geoscience | VOL 1 | OCTOBER 2008 | www.nature.com/naturegeoscience © 2008 Macmillan Publishers Limited. All rights reserved.
647
NEWS & VIEWS these extreme simulations because they are produced by two of the handful of coupled models that give relatively accurate simulations of the West African monsoon system. Biasutti and colleagues8 statistically analysed 21 of the GCM simulations from the IPCC archive, including the four outlier simulations from the two aforementioned models. They used simulations of the pre-industrial era, with only natural climate variability and no anthropogenic climate forcing, to build a linear regression model that relates modelled Sahel rainfall to modelled sea surface temperatures. They found that the GCMs capture the observed sensitivity of Sahel rainfall to sea surface temperature patterns — at least “in broad outline” — in both the pre-industrial era and in the 20th century under observed greenhouse gas increases, in agreement with a previous study10. But despite this dependence on sea surface temperatures in the past, the differences in the models’ projections of future rainfall are not explained by differences in their sea surface temperature projections. The relationship between sea surface temperatures and Sahel rainfall that holds for the 20th century and before does not seem to persist into the 21st century in the models. The authors therefore
suggest that simulated future changes in Sahel rainfall are controlled by different mechanisms than in the simulations of the past — for example, by a direct influence of the change in radiative forcing on precipitation, without mediation from sea surface temperatures. It is also possible that interactions between the land surface and the atmosphere are playing an important role. Biasutti and co-authors may be overly pessimistic about the level of agreement among the coupled GCMs as the large spread in the projected Sahel rainfall depends primarily on only two models (Fig. 1). These may be producing physically unrealistic results11, but dominating the statistics. Nonetheless, this mysterious result highlights the need for a deeper understanding of the West African monsoon system and the processes that cause it to vary on all time scales. The region is particularly challenging for accurate representation in GCMs for several reasons. One is that, in addition to the influence of global sea surface temperatures in the region, West Africa exhibits particularly strong land–atmosphere coupling12 and our ability to model these interactions is still relatively poor. Another challenge for GCMs arises because the model grids are still too coarse to fully resolve the tight
north–south gradients in temperature and moisture that characterize the region and that exert significant control on the flow and precipitation fields, or to resolve the all-important tropical convective processes. In order to get a better idea of what the future holds for the people of the Sahel, we will need to improve our basic understanding of precipitation and dynamical processes in West Africa. High-resolution modelling, physics-based analysis and a sophisticated and sustained observational network that is currently not in place are the means to this end. References 1. Dilley, J., Chen, R. S., Deichmann, U., Lerner-Lam, A. L. & Arnold, M. Natural Disaster Hotspots: A Global Risk Analysis (The World Bank, 2005). 2. Charney, J., Stone, P. H., Quirk, W. J. Science 187, 434–435 (1975). 3. Folland, C. K., Palmer, T. N., Parker, D. E. Nature 3210, 602–607 (1986). 4. Rowell, D. P. Quart. J. Roy. Meteor. Soc. 127, 1683–1706 (2001). 5. Hoerling, M., Hurrell, J., Eischeid, J. & Phillips, A. J. Climate 19, 3989–4008 (2006). 6. Paeth, H. & Hense, A. Climatic Change 65, 179–208 (2006). 7. Hagos, S. M. & Cook, K. H. J. Climate 21, 3797–3814 (2008). 8. Biasutti, M., Held, I. M., Sobel, A. H. & Giannini, A. J Climate, 21, 3471–3486 (2008). 9. Solomon, S. et al. (eds) Climate Change 2007: The Physical Science Basis — Contribution of Working Group 1 to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change (Cambridge Univ. Press, Cambridge, UK, 2007). 10. Joly, M., Voldoire, A., Douville, H., Terray, P. & Royer, J. F. Climate Dynamics 29, 1–20 (2007). 11. Cook, K. H. & Vizy, E. K. J. Climate 19, 3681–3703 (2006). 12. Koster, R. D. et al. J. Hydrometeor. 7, 590–610 (2006).
MINERAL PHYSICS
The spin deep within The electronic configuration of iron impurities in lower-mantle minerals influences their physical properties, but it is not well constrained. New studies suggest that ferrous iron in silicate phases exists mainly in an intermediate spin state.
Stephen Stackhouse is in the Department of Geological Sciences, University of Michigan, 2534 CC Little Bldg, 1100 North University Avenue, Ann Arbor, Michigan 48109-1005, USA. e-mail:
[email protected]
he Earth’s lower mantle, located between the depths of about 660 km to 2,890 km, is a region with a distinct mineralogical composition. The bulk of this region is made up of two mineral phases — silicate perovskite (~75%) and magnesiowüstite (~20%) — although in the lowermost few hundred kilometres, perovskite transforms into a denser
T 648
phase termed post-perovskite. Both silicate perovskite and post-perovskite are magnesium silicates that incorporate a significant amount of iron and aluminium into their structures as impurities through substitution. The effect of iron impurities on the properties of their host phases, and therefore the lower mantle as a whole, is a subject of great interest for mineral physicists. In this issue, Lin and colleagues1 (page 688) and McCammon and colleagues2 (page 684) present experiments aimed at understanding the electronic configuration of iron in lower-mantle silicate perovskite and post-perovskite and suggest that ferrous
iron mainly exists in an intermediate-spin state throughout this region. Establishing the electronic configuration of iron impurities in lower-mantle minerals is important because it can affect the radiative thermal conductivity of their host phase3, and this in turn could affect the structure and dynamics of the whole region4. Electronic configuration can also affect how iron is distributed between mineral phases5 and their elastic properties6. Depending on pressure and temperature conditions, iron impurities can adopt three different electronic configurations, known as high spin, intermediate spin and low spin.
nature geoscience | VOL 1 | OCTOBER 2008 | www.nature.com/naturegeoscience © 2008 Macmillan Publishers Limited. All rights reserved.
